On z/VM, using PER, you can see that MVCL is interruptible: issuing one for a 
big area (typically MB), you may see the same instruction traced multiple 
times. Kinda fun. At least that was true at some point; been a while since I've 
used PER in anger.

-----Original Message-----
From: IBM Mainframe Discussion List <[email protected]> On Behalf Of 
Attila Fogarasi
Sent: Friday, July 26, 2024 5:56 PM
To: [email protected]
Subject: Re: Does MVCDK move 'per byte' like MVC?

To answer "how can you tell", another TCB will see a partial move for MVCL 
while only the complete move is seen for MVC.  Rather important for some 
multi-tasking code.  On some older machines such as CMOS the partial move was 
on a cache line boundary, so by block but not predictable.  Don't know for 
current hardware, where cache management is much more sophisticated.
Made for some interesting bugs for pre-z.

On Sat, Jul 27, 2024 at 7:35 AM Paul Gilmartin < 
[email protected]> wrote:

> On Fri, 26 Jul 2024 16:03:40 -0500, Mike Schwab wrote:
>
> >MVC has another exemption.  The entire 256 bytes appear to move at 
> >once.
> >
>  How can you tell?  How might this "appear" different from moving 
> sequentially?  Was that true even on the earliest s/360?  Is that to 
> say that the move appears "block concurrent"?
>
> >   ... Vs MVCL appears to move 1 byte at a time with register updates.
>
> --
> gil
>
> ----------------------------------------------------------------------
> For IBM-MAIN subscribe / signoff / archive access instructions, send 
> email to [email protected] with the message: INFO IBM-MAIN
>

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions, send email to 
[email protected] with the message: INFO IBM-MAIN

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to